Prove adding SV will lower the risk, given the same E(r)

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
blueleaf
Posts: 134
Joined: Fri Dec 27, 2013 7:30 pm

Re: Prove adding SV will lower the risk, given the same E(r)

Post by blueleaf »

The key term here is Expected Return, and the wisdom of this strategy depends upon what you expect.

Larry has presented compelling evidence from the academic literature that small cap value stocks have had unusually high returns over a long period of time and around the world. If you you believe that pattern is a built in feature of capital markets that is very likely to persist in the future, then you can capitalize on it by owning more of this asset class. The choice then is whether to try for higher return for the same risk as the market portfolio, or to take a more typical level of expected return but lower your risk by owning more bonds. Note that if you do this, you're betting a big chunk of your expected return on your belief in a persistent scv premium. Not also that if you're making this bet, make sure you're really getting a lot of the smallest, cheapest stocks, as these are the ones that have really delivered in the past. As has been noted on this board many times vanguards vbr is pretty week tea as scv goes.

The other perspective is that the historic outperformance of scv will not necessarily continue. Either it was a historical fluke, or the conditions that supported it (poor liquidity, a lack of awareness by many market participants, a lack of investment vehicles) have changed. Publish a few papers about an anomaly and create some cheap liquid etfs, and here come the arbitraeurs, the smart money, the dumb money, the robo advisors, etc. Indeed, US small cap stocks have been on a tear of late, and now have scarily high valuations (though US scv is no more richly valued relative to small cap growth, that it has been historically.) If this story makes you doubt that scv will continue to deliver premium returns, then what you want is a market portfolio. This approach has the added advantage of being simpler, cheaper, and easier to explain to your loved ones.

Those are the facts. You get to choose which story is more compelling. You'll find strong cases made for both perspectives every day in this forum.

I can see validity I both sides. Like many here, I respect Larry's All In portfolio, but for my own circumstances I opt for a 'cowards tilt' . Mostly market, but a lean towards value. Keep in mind that this option only makes sense if you can do it very inexpensively and with minimal added possibility of behavioral error: the expected return premium for the typical boglehead coward's tilt is really pretty small, even if the historic patterns persist.

Many roads...
Topic Author
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by acegolfer »

Rodc wrote: I still do tilt small and value because for example I have an above average ability to handle "risk that shows up at bad times", and because if I factor in pensions, social security and low debt and low need for additional income I could actually rationally increase the risk I do hold even with the tilt. But I could do fine with TSM as well. And I don't think small or value is a free lunch.
Sounds like you are the Cochrane's "a wealthy investor with no other business or labor income" who are less affected in times of financial distress. According to Cochrane (1999), you are an ideal candidate to tilt to value for extra premium.

The biggest issue I personally have with SV proponents is that they sound like SV tilting only has benefits. They rarely mention the cons of tilting, which sounds like SV is a free lunch. When something sounds free, I step back and try to find out the pitfalls. In this case, S and V increase risks other than stdev, or tail risk. In addition, there's no proof that SV are multi-factor efficient. Without knowing what these risks are and whether SV is efficient, I'll be conservative and stick with MKT, which we know is multi-factor efficient.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

hafius
Re debt premiums, the term premium has been rewarded, at least out to intermediate term, it's the credit/default premium that has been poorly rewarded
Larry
Tamales
Posts: 1644
Joined: Sat Jul 05, 2014 10:47 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by Tamales »

MrMatt2532 wrote:I just ran a bunch of numbers using the FF data from 1927-2013. Figured it might be useful for the discussion. Generally, you can see, the data shows one could have owned a mix of small and value, decreased beta, and still achieved the same expected return as the market. I computed the sharpe ratio and a few others to assess the risk adjusted return. Omega and sortino ratio are specifically formulated to handle effects besides just return and standard deviation of returns (i.e. other features of the distribution). As you can see, the skew and kurtosis numbers are favorable for small value tilting as compared with the market. Let me know if there are any questions.
MrMatt, this looks potentially interesting, and I hope you don't take offense that I can't get very far in deciphering it. Would you mind going into a bit more detail, including all the abbreviations you're using? Or maybe even better if you could post the actual spreadsheet? Also wondering, are you using annual data (I'm guessing that's all you can get in going back to the 1920's, and I feel annual rather than at least monthly will miss a lot of "risk events" both up and down)
Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by Rodc »

acegolfer wrote:
Rodc wrote: I still do tilt small and value because for example I have an above average ability to handle "risk that shows up at bad times", and because if I factor in pensions, social security and low debt and low need for additional income I could actually rationally increase the risk I do hold even with the tilt. But I could do fine with TSM as well. And I don't think small or value is a free lunch.
Sounds like you are the Cochrane's "a wealthy investor with no other business or labor income" who are less affected in times of financial distress. According to Cochrane (1999), you are an ideal candidate to tilt to value for extra premium.

The biggest issue I personally have with SV proponents is that they sound like SV tilting only has benefits. They rarely mention the cons of tilting, which sounds like SV is a free lunch. When something sounds free, I step back and try to find out the pitfalls. In this case, S and V increase risks other than stdev, or tail risk. In addition, there's no proof that SV are multi-factor efficient. Without knowing what these risks are and whether SV is efficient, I'll be conservative and stick with MKT, which we know is multi-factor efficient.
"a wealthy investor with no other business or labor income"

Not quite there, but closing in...

Not so much wealthy as upper middle class with simple tastes, low debts, a much more stable job than most, and close to retirement which will bring two pensions.

I agree with the rest of what you wrote. If you only look at SD as risk and use only the FF data, SVC looks fabulous. I made the same mistake in the beginning. But that completely misses the point of why the factor models were developed.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.
cowboysFan
Posts: 290
Joined: Sat Jan 25, 2014 12:46 pm

Re: Prove adding SV will lower the risk, given the same E(r)

Post by cowboysFan »

larryswedroe wrote:cowboys
Repeating the error of isolation thinking. The higher risk-adjusted returns have come not from adding SV but from adding SV and lowering beta at the same time.
Second issue is you are assuming that the advisor adds no value beyond access if only adding cost
Larry
I didn't include the advisor fees, only the 0.47% ER of the fund itself. I did look at what happened when you added SV and lowered beta.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

cowboys,
Well the fund has virtually matched its benchmark index since inception, underperforming by just 8bp---and ER cost has come down over time I believe and they only started using neg MOM screens in 2003,
I ran the fund vs the FF SV ex utilities index for last 10 years when data available, ending May 2014, fund outperformed by 8.93 to 8.80. Of course that's not statistically significant, or would not think so, but it is 10 years of live data and includes lower ER of fund and MOM screen,
Take it for what it's worth
Larry
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by MrMatt2532 »

Tamales wrote: MrMatt, this looks potentially interesting, and I hope you don't take offense that I can't get very far in deciphering it. Would you mind going into a bit more detail, including all the abbreviations you're using? Or maybe even better if you could post the actual spreadsheet? Also wondering, are you using annual data (I'm guessing that's all you can get in going back to the 1920's, and I feel annual rather than at least monthly will miss a lot of "risk events" both up and down)
No problem. Here's a quick abbreviation list:
b3 = market risk exposure, bs = small stock risk exposure, bv=value risk exposure, rf=risk free rate
average=average of the returns, std=standard deviation of the returns, skew=skewness of the returns, kurt=kurtosis of the returns
sharpe=sharpe ratio, M2=Modigliani risk-adjusted performance, sortino=sortino ratio, omega=omega ratio
Really its just number crunching and using the optimization functionality in excel.
The major point is that tilting to SV did improve any risk adjusted metric that you might throw at it. Additionally, the data does show that you were able to cut fat tails and specifically improve (increase) the skew.

I did use annual data, however I plugged in the monthly data and converted it to annual units for you. The results are below:
Image
Tamales
Posts: 1644
Joined: Sat Jul 05, 2014 10:47 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by Tamales »

MrMatt2532 wrote: I did use annual data, however I plugged in the monthly data and converted it to annual units for you. The results are below:
Thanks Matt. Some of those changed quite a lot using the monthly data. Modigliani is one I've not heard of before. Are you saying all those ratios are native calculations in Excel, or do you have some add-in tool pack?
I'll have to look some of those stats up to see if I can figure out how to interpret the numbers. I gather than most of them should be judged in comparison to the other things and don't reveal much as a stand-alone number.
cowboysFan
Posts: 290
Joined: Sat Jan 25, 2014 12:46 pm

Re: Prove adding SV will lower the risk, given the same E(r)

Post by cowboysFan »

larryswedroe wrote:cowboys,
Well the fund has virtually matched its benchmark index since inception, underperforming by just 8bp---and ER cost has come down over time I believe and they only started using neg MOM screens in 2003,
I ran the fund vs the FF SV ex utilities index for last 10 years when data available, ending May 2014, fund outperformed by 8.93 to 8.80. Of course that's not statistically significant, or would not think so, but it is 10 years of live data and includes lower ER of fund and MOM screen,
Take it for what it's worth
Larry
That's interesting, but I''m not sure what to make of it. If you assume 10 basis points for trading costs and use their latest ER of 0.52%, DFSVX has beaten their index by about 0.75% a year over the past 10 years before costs. How do you follow an index and beat it by 0.75% a year?
rca1824
Posts: 719
Joined: Tue Jun 03, 2014 9:33 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by rca1824 »

My proof is simple: the expected returns of SV is higher than TSM, and the correlation is less than 1. SV has a unique factor exposure, just as bonds exposure you to credit risk and interest rate risk. Risks are rewarded, and it's better to diversify across as many unique risks are possible to reduce total portfolio volatility. For the same reason you hold TSM instead of one company, holding SV in addition to TSM exposes you to more than just one fundamental factor. Why expose yourself to only market risk when you can also be exposed to small and value risks?

Even if you consider small and value to be more volatile, there is still some optimal positive quantity of each to hold.

Here's a good post by Economist Maniw http://gregmankiw.blogspot.com/2013/07/on-au.html on why it would be rational to hold gold in additional to stocks and bonds, under certain assumptions. The general point is that it's optimal to diversify across as many uncorrelated, positive-expectation-return assets as is practical and possible.
Monthly or yearly movements of stocks are often erratic and not indicative of changes in intrinsic value. Over time, however, stock prices and intrinsic value almost invariably converge. ~ WB
steve_14
Posts: 1507
Joined: Wed Jun 20, 2012 12:05 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by steve_14 »

rca1824 wrote:My proof is simple: the expected returns of SV is higher than TSM, and the correlation is less than 1. SV has a unique factor exposure, just as bonds exposure you to credit risk and interest rate risk.
All also true of small growth, which is clearly as risky, if not more so, than SV in real fund results. Its P/E is currently 42! Are you sure that's low risk? And while small has a higher expected return, it also has higher risk, so no free lunch there. The "correlation less than 1" argument doesn't hold water either. If I overweight stocks west of the Mississippi river, and you overweight stocks east of it, we can't both rebalance to a higher Sharpe ratio.
Topic Author
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by acegolfer »

rca1824 wrote:My proof is simple: the expected returns of SV is higher than TSM, and the correlation is less than 1. SV has a unique factor exposure, just as bonds exposure you to credit risk and interest rate risk. Risks are rewarded, and it's better to diversify across as many unique risks are possible to reduce total portfolio volatility. For the same reason you hold TSM instead of one company, holding SV in addition to TSM exposes you to more than just one fundamental factor. Why expose yourself to only market risk when you can also be exposed to small and value risks?

Even if you consider small and value to be more volatile, there is still some optimal positive quantity of each to hold.

Here's a good post by Economist Maniw http://gregmankiw.blogspot.com/2013/07/on-au.html on why it would be rational to hold gold in additional to stocks and bonds, under certain assumptions. The general point is that it's optimal to diversify across as many uncorrelated, positive-expectation-return assets as is practical and possible.
Thanks for you inputs. Here are my comments to each statement.
1. SV with higher E(r) and not perfectly correlated with MKT does not necessarily mean one can both increase E(r) and lower stdev at the same time. You must prove SV stdev is also smaller than MKT. In case you didn't read OP, the challenge of this thread is to prove without relying on just simple statistics.
2. You stated SV has unique factor exposure, which means when adding SV to MKT, it will increase these risks (not stdev). That's hardly a proof of lowering the risk, given the same E(r), unless you define risk = stdev.
3. In a multi-factor model, MKT addresses all risks and hence is multi-factor efficient. If you assume MKT minimizes only 1 risk (stdev) given E(R), then MKT must be mean-variance efficient, which we all know is false. MKT being multi-factor efficient means it is already diversified across many unique risks at the expense of higher stdev.
4. When you say "there is still some optimal positive quantity of each to hold", I assume it's at the aggregate level (not for individual). At the aggregate, MKT already holds a positive quantity of each. OTOH, if you are saying MKT must hold more of S and V, then the MKT is not in equilibrium and MKT will tilt to S and V until equilibrium.
5. Mankiw said adding gold (which is not stock/bond) to stocks/bonds is a diversification. If you say adding SV (which is already included in MKT) to MKT is a diversification, then you should also say adding APPL to TSM is a diversification.

If I'm being too critical to you, my apologies. If you can't take this criticisms, I'll not comment on your posts any more.
Topic Author
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by acegolfer »

Kevin M wrote:
acegolfer wrote:I didn't mention this because this thread is not about where to tilt.
But the title of the thread mentions "adding SV", which is what we mean by tilting to SV. If this thread is not about tilting to SV, then what is it about?

Kevin
Kevin,

I thought about this for a few days and now have an answer. The difficulty that I had earlier was how to prove SV is also multi-factor efficient. But I realized that multi-factor efficiency is not necessary in order to prove adding SV will increase E(r) and lower stdev.

In my earlier post, I provided a proof showing that it's possible to increase E(r) and lower stdev by adding another multi-factor efficient portfolio. Even if SV is not efficient, as long as SV has higher E(r), lower stdev, then adding SV to MKT will increase E(r), lower stdev at the expense of higher beta (measure of other risks, not CAPM beta).

Most proponents of SV tilting showed that SV has higher E(r) and lower stdev using sample statistics with different periods to show persistence. However, I was not convinced that these results extends to true parameters because SV sounded a free lunch, which cannot happen in a rational world. I'm now convinced that SV has higher E(r) and smaller stdev than MKT, which is possible in a multi-factor model. A note of caution to readers: having lower stdev doesn't mean lower risk because by adding SV, you are increasing other risks. So try to understand what sacrifice you are making when you lower a certain risk. SV is not a free lunch.

Note: I used others' sample statistics to prove that adding SV can increase E(r) and lower stdev. But I didn't rely only on those stats but also provided a theoretical justification why it's possible.
Last edited by acegolfer on Sat Aug 02, 2014 10:03 am, edited 1 time in total.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

cowboys
First, you can also find periods of underperformance
But second, you do it by doing the things I wrote about in the negatives of indexing piece. There are some negatives which can be minimized or avoided. You also earn securities lending fees. You patiently trade, buying blocks sometimes at discounts to market (indexers cannot do or risk tracking error), you screen out negative momentum, and so on. I certainly would not expect the fund to do much more than about match the index after costs, or come close, but pure indexing has flaws which can be addressed with intelligent design if willing to accept what should be random tracking error. Investors should not care about matching some index, but about having exposure to the factors they want in most efficient way--the price of that is random tracking error.

Larry
Topic Author
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by acegolfer »

larryswedroe wrote:cowboys
First, you can also find periods of underperformance
But second, you do it by doing the things I wrote about in the negatives of indexing piece. There are some negatives which can be minimized or avoided. You also earn securities lending fees. You patiently trade, buying blocks sometimes at discounts to market (indexers cannot do or risk tracking error), you screen out negative momentum, and so on. I certainly would not expect the fund to do much more than about match the index after costs, or come close, but pure indexing has flaws which can be addressed with intelligent design if willing to accept what should be random tracking error. Investors should not care about matching some index, but about having exposure to the factors they want in most efficient way--the price of that is random tracking error.

Larry
Larry or any SV proponents,

I now understand the pros of your strategy. But you only mentioned the cons of indexing without mentioning the cons of tilting. If you think there are no cons to SV tilting, then we can start a new debate perhaps in another thread. But if you think there are cons but intentionally not addressing them, then it's unfair. If you have mentioned the cons in another thread, can you direct me? I couldn't find it by searching.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

Ace

First I have never RECOMMENDED a tilt, I have recommended investors CONSIDER tilting given the evidence and the trade offs. I have also clearly stated there is no one right portfolio, just one right for each person with the most important point being the discipline to stick with whatever the strategy/allocation is. And never investing in strategy don't fully understand the nature of the risks, the pros and cons.

I clearly mentioned and even warned that the biggest con to tilting is the ability to stay the course even if the strategy appears not to be working----when you have years like 1998 and big tracking error and underperformance. And of course there is the clear risk of no risk premium ex-post. And I also clearly note that one trades off cutting the left tail for reduced right tail opportunity, so stating I have not mentioned the cons is just wrong. This is an issue we/I spend lots of time on discussing before ever implementing any asset allocation. And we have many clients with varying degrees of tilts, with only small percentage adopting the "LP" approach.

Having said that if you look at the data and the logic, for those that have already won the game, little to no need to take risk there are really no negatives other than discipline question because it's very hard to even imagine period of SV having negative returns without beta doing do so. There are no such periods and none I could see likely. Now if you give up the good right tail you don't care, the strategy worked and you new that risk and you accepted it---the price of left tail insurance---would be like complaining you wasted your insurance premiums because risk did not show up.

Larry
Topic Author
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by acegolfer »

larryswedroe wrote:Ace
I clearly mentioned and even warned that the biggest con to tilting is the ability to stay the course even if the strategy appears not to be working----when you have years like 1998 and big tracking error and underperformance. And of course there is the clear risk of no risk premium ex-post. And I also clearly note that one trades off cutting the left tail for reduced right tail opportunity, so stating I have not mentioned the cons is just wrong.

Having said that if you look at the data and the logic, for those that have already won the game, little to no need to take risk there are really no negatives other than discipline question because it's very hard to even imagine period of SV having negative returns without beta doing do so. There are no such periods and none I could see likely. Now if you give up the good right tail you don't care, the strategy worked and you new that risk and you accepted it---the price of left tail insurance---would be like complaining you wasted your insurance premiums because risk did not show up.

Larry
Again, thanks for the quick reply. Can you address the following issues of SV and your pro argument that I can think of?

1. SV may have low beta (market risk) but high E(r). But as long as size and value are risk factors, SV will carry higher risks in these dimensions. IMO, not addressing these risks is totally unfair. What you mentioned above are not the risks in the context of risk factors.

2. Related to the above, according to Cochrane (1999), value stocks tend to have worse performance in times of financial distress than other stocks. (I think you also mentioned this in another thread.) So even if SV has low beta, when economy crashes, return of SV may suffer more, which can increase the tail risk rather than reducing it. By the way, 0 covariance (zero beta) between SV and MKT doesn't mean there's no relationship between the 2. (Think of a U shape relationship.)

3. In a multi-factor world, the market portfolio is multi-factor efficient but not mean-variance (1 risk, stdev, world) efficient. This means MKT doesn't only consider stdev but also other risk as well. In other words, MKT already diversified many risk factors. Using risk factor diversification to promote SV assumes MKT doesn't diversify between risk factors, which is not true. If you are going to argue that MKT has a MKT loading of 1 and 0 for SMB and HML so it's not risk factor diversified, then I can make the same argument that small has SMB loading of 1 and 0 for other factors so small is not risk factor diversified. Next, if you are going to claim that SV is not SMB or HML so it's diversified, then explain why SV is multi-factor efficient. To sum, I argue that MKT is not only efficient but also has factor diversification benefit that you emphasized.

4. A fundamental truth of investing is that the the average investor must hold the market portfolio. Investors in your group tilt one way. Can you explain why some tilt to the opposite direction? Your arguments for SV (namely left tail risk) should apply to all investors. But why don't these arguments not apply to some and they run away from SV?

[OT comment removed by admin LadyGeek]
Jebediah
Posts: 805
Joined: Tue Aug 28, 2012 9:19 pm
Location: Austin TX

Re: Prove adding SV will lower the risk, given the same E(r)

Post by Jebediah »

acegolfer wrote: Again, thanks for the quick reply. Can you address the following issues of SV and your pro argument that I can think of?

1. SV may have low beta (market risk) but high E(r). But as long as size and value are risk factors, SV will carry higher risks in these dimensions. IMO, not addressing these risks is totally unfair. What you mentioned above are not the risks in the context of risk factors.

Taking concentrated risk means ability to lower overall quantity of risk taken so you can use that money for other things (bonds or whatever)

2. Related to the above, according to Cochrane (1999), value stocks tend to have worse performance in times of financial distress than other stocks. (I think you also mentioned this in another thread.) So even if SV has low beta, when economy crashes, return of SV may suffer more, which can increase the tail risk rather than reducing it. By the way, 0 covariance (zero beta) between SV and MKT doesn't mean there's no relationship between the 2. (Think of a U shape relationship.)

Yes SV is more exposed to deflation/recession risk. But again, see #1. As Cochrane points out, investors should consider the specifics of their situation when deciding what kinds of risk to allocate to.

3. In a multi-factor world, the market portfolio is multi-factor efficient but not mean-variance (1 risk, stdev, world) efficient. This means MKT doesn't only consider stdev but also other risk as well. In other words, MKT already diversified many risk factors. Using risk factor diversification to promote SV assumes MKT doesn't diversify between risk factors, which is not true. If you are going to argue that MKT has a MKT loading of 1 and 0 for SMB and HML so it's not risk factor diversified, then I can make the same argument that small has SMB loading of 1 and 0 for other factors so small is not risk factor diversified. Next, if you are going to claim that SV is not SMB or HML so it's diversified, then explain why SV is multi-factor efficient. To sum, I argue that MKT is not only efficient but also has factor diversification benefit that you emphasized.

MKT is not risk factor diversified-- it negates its size and value exposure by being long growth and large (recall factors are long/short). A port of long SV stocks loads on all 3 factors, mostly on beta unless it is really micro and deep value.

4. A fundamental truth of investing is that the the average investor must hold the market portfolio. Investors in your group tilt one way. Can you explain why some tilt to the opposite direction? Your arguments for SV (namely left tail risk) should apply to all investors. But why don't these arguments not apply to some and they run away from SV?

Lots of reasons for people to take the other side of the trade. Recession/deflation sensitive investors might not want much size/value exposure. People like small growth because of the lotterly-like features. "Buy good American growth funds" is a Dave Ramsey cliche that "sounds right" to the everyman investor (and a lot of wall street as well). Some smart people on these threads eschew it because it sounds too good to be true. etc...

larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

First, don't know where got idea about SV having low beta. It's the "LP" that has the low beta allowed by tilting to SV. So reducing beta at same time. But SV doesn't have low beta. In fact both small and value stocks over the long term have high betas.

Second, Re Cochrane, exactly why IMO it's a risk factor and why premium should be large (but likely too large so some behavioral story, especially related to overpaying for small growth in particular) And this is why you should consider LOWER beta at same time you tilt more. Remember I am not arguing to tilt without lowering beta unless have ability, willingness and need to take more risk--then okay as at least diversifying sources of returns away from just beta. But then should have high stability of labor capital. And better way to raise expected returns IMO than margin or raising beta exposure which just adds more of the same risk. Especially true for those constrained by policy limits.

Third, it's the PORTFOLIO's loadings that matter, not the individual asset class. But small doesn't have loading of 0 on other factors, it loads on beta, so don't know how you argue that. AS to market loading, I'm not arguing anything, just stating facts. In multi-factor world if you believe market efficient than prices are right and only a choice of loading based on preferences. I don't buy it, as market not perfectly efficient, though it's good model to start with. IMO many behavioral anomalies around that "prove" market not efficient even in multi-factor world. Even Fama would agree with that. Even he would cite small growth anomaly and momentum as two huge challenges---Note DFA has long excluded extreme small growth stocks and other "lottery tickets" based on the evidence.

Fourth, obviously not all can tilt to 2% of market. There are many behavioral reasons why many won't and also simply lack of knowledge of finance and of course you have those that don't believe the evidence justifies a tilt. And then there are those whose labor capital should cause them to think about not tilting.


[Response to OT comment removed by admin LadyGeek]
Larry
Last edited by larryswedroe on Sat Aug 02, 2014 12:46 pm, edited 1 time in total.
User avatar
LadyGeek
Site Admin
Posts: 95696
Joined: Sat Dec 20, 2008 4:34 pm
Location: Philadelphia
Contact:

Re: Prove adding SV will lower the risk, given the same E(r)

Post by LadyGeek »

I removed an off-topic comment and its response. As a reminder, see: Forum Policy
We expect this forum to be a place where people can feel comfortable asking questions and where debates and discussions are conducted in civil tones.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
Topic Author
acegolfer
Posts: 3029
Joined: Tue Aug 25, 2009 9:40 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by acegolfer »

larryswedroe wrote:First, don't know where got idea about SV having low beta. It's the "LP" that has the low beta allowed by tilting to SV. So reducing beta at same time. But SV doesn't have low beta. In fact both small and value stocks over the long term have high betas.

Second, Re Cochrane, exactly why IMO it's a risk factor and why premium should be large (but likely too large so some behavioral story, especially related to overpaying for small growth in particular) And this is why you should consider LOWER beta at same time you tilt more. Remember I am not arguing to tilt without lowering beta unless have ability, willingness and need to take more risk--then okay as at least diversifying sources of returns away from just beta. But then should have high stability of labor capital. And better way to raise expected returns IMO than margin. Especially true for those constrained by policy limits.

Third, it's the PORTFOLIO's loadings that matter, not the individual asset class. But small doesn't have loading of 0 on other factors, it loads on beta, so don't know how you argue that. AS to market loading, I'm not arguing anything, just stating facts. In multi-factor world if you believe market efficient than prices are right and only a choice of loading based on preferences. I don't buy it, as market not perfectly efficient, though it's good model to start with. IMO many behavioral anomalies around that "prove" market not efficient even in multi-factor world. Even Fama would agree with that. Even he would cite small growth anomaly and momentum as two huge challenges---Note DFA has long excluded extreme small growth stocks and other "lottery tickets" based on the evidence.

Fourth, obviously not all can tilt to 2% of market. There are many behavioral reasons why many won't and also simply lack of knowledge of finance and of course you have those that don't believe the evidence justifies a tilt. And then there are those whose labor capital should cause them to think about not tilting.


[Response to OT comment removed by admin LadyGeek]
Larry
1. My mistake. Whenever you say tilting to SV lowers beta, I assumed SV has zero beta. In addition, in pg 1, someone posted "small" and "value" loadings. Beta for small and value were 0s. So I thought "small" is SMB. Apparently, his "small" and your "small" are different.

2. If SV is a risk, as you acknowledged, I think you should mention this risk when talking about the cons of SV tilting.

3. Once again, I assumed "small" meant SMB because of another poster. Again, my mistake. OTOH, Fama does say MKT is efficient. (there's another thread quoting his interview. but he does state that MOM is the biggest challenge to EMH.) If you claim that MKT is not efficient, can you then claim small is efficient? Or efficiency doesn't matter at all in portfolio selection? More importantly, having anomalies in a FF-3 factor model does not mean EMH is false. Testing an asset pricing model is always a joint hypothesis of EMH and the model. Even if we found anomalies to FF-3 model, we can't tell whether the market is inefficient or the model is wrong. In this case, IMO, FF-3 factor model is not sufficient to explain the E(r).

4. I didn't ask why some don't tilt to SV. I asked why some tilt the opposite direction so that the average investor holds MKT. Your first 2 answered why ppl don't tilt. But your last argument answered my question.

I understand why some people introduce behavioral reasons to explain phenomena or anomalies. But I try to understand these using rationality.

And most importantly, my deepest apology with my last comment. I didn't mean it to be snarky or offensive at all. I was worried that my comments and questions were too critical. If you are okay with my comments and criticisms, then I'm relieved. I greatly appreciate you for continuing this discussion.
MrMatt2532
Posts: 454
Joined: Sun Mar 15, 2009 11:58 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by MrMatt2532 »

Tamales wrote: Thanks Matt. Some of those changed quite a lot using the monthly data. Modigliani is one I've not heard of before. Are you saying all those ratios are native calculations in Excel, or do you have some add-in tool pack?
I'll have to look some of those stats up to see if I can figure out how to interpret the numbers. I gather than most of them should be judged in comparison to the other things and don't reveal much as a stand-alone number.
Yes, I was a little bit surprised at some of the differences between the monthly vs yearly results. I had figured with 80 years of data it wouldn't be too different. Nonetheless, the main takeaways were the same I would say.

The Modigliani metric is actually a reworking of the sharpe ratio made to be more easily interpreted. Also, no, none of them are native to excel, they are just pretty straightforward to calculate. For example, if you have a list of (return minus risk free rate), you can just take the average divided by the standard deviation of that list to get the sharpe ratio. And yes I would absolutely agree they need to be judged in comparison with each other.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

Ace
Re, never said tilting to SV lowers beta. I stated that there are two ways to use the premium when tilting, raise risk and expected return or keep expected return the same by lowering beta exposure.
Could not have been more clear on that. And typical SV portfolio has beta of around 1. As do most diversified stock portfolios.

BTW- to be clear SmB is return of small minus return of large, annual average. Like all factors long short. But a small portfolio while being exposed to small will also load heavily, like 1 or more, on Beta.

Re risk of SV --I mention it basically all the time, it's in all my books, in many blog posts written, etc. I have even posted lists of over 20 papers on subject. How much more can I do?

What Fama says "publicly" and the reality are sometimes very different. Publicly likely talking in generalties. As head of research at DFA he excluded extreme small growth stock and other anomalies as well as began incorporating MOM.

As to small, IMO small not efficient because small growth has been mispriced. Been called the black hole in fact. As to anomalies, we know the EMH is false, not the model which isn't perfect either, as no models re. There is no model to explain anomalies like small growth, iPOs, etc---except preference for skewness--which is what Fama wrote paper on btw. For those investors SG is efficient, and even rational, but not for what economists would call "rational"--meaning demand premium for higher risk. Personally IMO there are those so hung up on perfect market efficiency that they are blinded to whatever information and evidence there is.

The problem with understanding behavior and anomalies is that they cannot be explained by what economists would call rational behavior. That's why they are anomalies.

And apology accepted. I never take QUESTIONS as you asked them that way. I took them as simple questions/debate. It was the "big man" comment that got me. There are no bad questions, just ones you don't ask or ask in offensive way----which often you can tell.

And always happy as others have noted to take PMs, answer everyone of them.

Larry
User avatar
Kevin M
Posts: 15789
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Prove adding SV will lower the risk, given the same E(r)

Post by Kevin M »

Can't remember if it's been shared in this thread yet, but here's a paper by Fama and French on the market, SmB and HmL premiums and their volatility:

us.dimensional.com/pdf/Volatility_And_Premiums.pdf

I think this graph and the similar one for SmB volatility are helpful:

Image

Kevin
If I make a calculation error, #Cruncher probably will let me know.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

Kevin
yes the sd of the premiums are multiples of the premiums themselves, and that's one reason there is a big TE risk problem
Larry
User avatar
Kevin M
Posts: 15789
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Prove adding SV will lower the risk, given the same E(r)

Post by Kevin M »

Another observation from these graphs is that there is no obvious, consistent reduction of the small and value premiums since their "discovery".

Kevin
If I make a calculation error, #Cruncher probably will let me know.
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

Kevin
The small value premium has been pretty much the same in the post and pre FF 1992 paper periods as has been pointed out many times. And the series on "overgrazing" by Claude Erb stated SV was the one premium that there was no evidence of shrinkage.
Larry
rca1824
Posts: 719
Joined: Tue Jun 03, 2014 9:33 am

Re: Prove adding SV will lower the risk, given the same E(r)

Post by rca1824 »

acegolfer wrote:
rca1824 wrote:My proof is simple: the expected returns of SV is higher than TSM, and the correlation is less than 1. SV has a unique factor exposure, just as bonds exposure you to credit risk and interest rate risk. Risks are rewarded, and it's better to diversify across as many unique risks are possible to reduce total portfolio volatility. For the same reason you hold TSM instead of one company, holding SV in addition to TSM exposes you to more than just one fundamental factor. Why expose yourself to only market risk when you can also be exposed to small and value risks?

Even if you consider small and value to be more volatile, there is still some optimal positive quantity of each to hold.

Here's a good post by Economist Maniw http://gregmankiw.blogspot.com/2013/07/on-au.html on why it would be rational to hold gold in additional to stocks and bonds, under certain assumptions. The general point is that it's optimal to diversify across as many uncorrelated, positive-expectation-return assets as is practical and possible.
Thanks for you inputs. Here are my comments to each statement.
1. SV with higher E(r) and not perfectly correlated with MKT does not necessarily mean one can both increase E(r) and lower stdev at the same time. You must prove SV stdev is also smaller than MKT. In case you didn't read OP, the challenge of this thread is to prove without relying on just simple statistics.
2. You stated SV has unique factor exposure, which means when adding SV to MKT, it will increase these risks (not stdev). That's hardly a proof of lowering the risk, given the same E(r), unless you define risk = stdev.
3. In a multi-factor model, MKT addresses all risks and hence is multi-factor efficient. If you assume MKT minimizes only 1 risk (stdev) given E(R), then MKT must be mean-variance efficient, which we all know is false. MKT being multi-factor efficient means it is already diversified across many unique risks at the expense of higher stdev.
4. When you say "there is still some optimal positive quantity of each to hold", I assume it's at the aggregate level (not for individual). At the aggregate, MKT already holds a positive quantity of each. OTOH, if you are saying MKT must hold more of S and V, then the MKT is not in equilibrium and MKT will tilt to S and V until equilibrium.
5. Mankiw said adding gold (which is not stock/bond) to stocks/bonds is a diversification. If you say adding SV (which is already included in MKT) to MKT is a diversification, then you should also say adding APPL to TSM is a diversification.

If I'm being too critical to you, my apologies. If you can't take this criticisms, I'll not comment on your posts any more.
It just comes down to whether or not you believe that SV introduces new, unique risk factors that are not present in TSM. Adding APPL to TSM isn't a good example because APPL is a single company and there's no reason to believe that it's past performance is a predictor of future performance. SV, on the other hand, is categorically different, just as equities are categorically different from bonds and real estate and gold.

And for the some reason that adding some stocks to a 100% bond portfolio lowers portfolio risk, adding some SV to an equity portfolio lowers risk as well because of the uncorrelation. The more diversified your risk factors are, the lower your total portfolio risk. And because SV E[R] is higher than TSM, you can then reduce your overall equity exposure and increase bonds to reach risk parity but still have higher E[R], or you can reduce equity further until you have parity with E[R] but at less risk.

I don't see this as implying that MKT should, in equilibrium, tilt towards S and V until the premium disappears. The individually optimally portfolio need not look like the aggregate optimal portfolio. The aggregate can have systematic biases causing distortions that, if an individual can recognize, can profit on. A classic example is how stock prices crashed in 2008, only to recover quickly thereafter. Was the price crash an efficient and accurate proxy for the true underlying value of the stock market? Of course not--it was driven by animal spirits and the ignorant masses panicking and pulling their money out. But the contrarian investor who recognizes that the masses are irrational and who can bet against them then can easily profit. This is how many famous investors have made their fortune.
Monthly or yearly movements of stocks are often erratic and not indicative of changes in intrinsic value. Over time, however, stock prices and intrinsic value almost invariably converge. ~ WB
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: Prove adding SV will lower the risk, given the same E(r)

Post by larryswedroe »

RCA
I don't agree with this at all
Was the price crash an efficient and accurate proxy for the true underlying value of the stock market? Of course not--it was driven by animal spirits and the ignorant masses panicking and pulling their money out
Easily said in hindsight. No one knew that the Fed's policy's and other actions would succeed. It certainly might have turned out differently.
Larry
Post Reply